Prepare Oracle Volume

Due to legal reasons, the Oracle driver and slqplus client cannot be shipped with the components themselves. To make use of the automatic migration provided by the Managed Database pattern, a Kubernetes persistent volume has to be prepared in the namespace of the used component that contains these dependencies. This is currently only supported for nevisIDM.

This is done by applying the following yaml, which creates a persistent volume, and a job that downloads the needed dependencies from Oracle.

oracle-volume.yaml

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: oracle-volume
spec:
  accessModes:
  - ReadWriteMany
  resources:
    requests:
      storage: 1Gi
  storageClassName: azurefile
  volumeMode: Filesystem
---
apiVersion: batch/v1
kind: Job
metadata:
  name: prepare-oracle
spec:
  backoffLimit: 1
  completions: 1
  parallelism: 1
  template:
    spec:
      securityContext:
        fsGroup: 1001
      volumes:
      - name: oracle-volume
        persistentVolumeClaim:
          claimName: oracle-volume
      containers:
      - command: ["/bin/sh", "-c"]
        args:
        - mkdir -p /tmp/oracle;
          curl --fail https://download.oracle.com/otn_software/linux/instantclient/1915000/instantclient-basiclite-linux.x64-19.15.0.0.0dbru-2.zip -o /tmp/oracle/basic.zip;
          curl --fail https://download.oracle.com/otn_software/linux/instantclient/1915000/instantclient-sqlplus-linux.x64-19.15.0.0.0dbru-2.zip -o /tmp/oracle/sqlplus.zip;
          unzip -o -d `dirname /tmp/oracle/basic.zip` /tmp/oracle/basic.zip;
          unzip -o -d `dirname /tmp/oracle/sqlplus.zip` /tmp/oracle/sqlplus.zip;
          cp -dR /tmp/oracle/instantclient_19_15/* /var/opt/oracle/;
        volumeMounts:
        - mountPath: "/var/opt/oracle"
          name: oracle-volume
        image: <registry>/nevis/nevis-ubi-tools:1.3.0
        imagePullPolicy: Always
        name: schema-init
        resources:
          limits:
            cpu: "1"
            memory: 1000Mi
          requests:
            cpu: 20m
            memory: 200Mi
        terminationMessagePath: /dev/termination-log
        terminationMessagePolicy: File
      dnsPolicy: ClusterFirst
      restartPolicy: Never
      schedulerName: default-scheduler
      securityContext: {}
      terminationGracePeriodSeconds: 30

kubeclt apply -f oracle-volume.yaml -n <component-namespace>

Replace <registry> with the used registry and azurefile with a different StorageClass that supports the ReadWriteMany access mode such as longhorn or nfs in case azurefile is not available on the cluster.